Many epidemiological studies (Kajantie et al., 2009; Wu et al., 2009, 2011; Davis et al., 2012a,b; Lawlor et al., 2012) indicate that preeclampsia is associated with long-term adverse outcomes in the offspring. The majority of studies (Palti and Rothschild, 1989; Seidman et al., 1991; Tenhola et al., 2003, 2006; Vatten et al., 2003; Swarup et al., 2005; Hiller et al., 2007; Oglaend et al., 2009; Kvehaugen et al., 2010; Lazdam et al., 2010; Palmsten et al., 2010; Lawlor et al., 2012), but not all (Ounsted et al., 1983; Jayet et al., 2010; Belfort et al., 2012; Lawlor et al., 2012) report that children and adolescents who were exposed to preeclampsia or hypertension in pregnancy exhibit higher systolic and diastolic blood pressure compared with non-exposed children or adolescents. These studies were reviewed in a recent meta-analysis (Davis et al., 2012a), which included individuals aged 4–30 years, born at term from preeclamptic pregnancies. This meta-analysis concluded that offspring born from preeclamptic women had ~2 mm Hg greater systolic and ~1.3 mm Hg greater diastolic blood pressure than individuals born from normotensive pregnancies. Interestingly, according to their prediction, “if the 2.4 mmHg difference in systolic blood pressure tracks into adulthood (Chen and Wang, 2008), this difference would be associated with an ~8% increased risk of mortality from ischemic heart disease and 12% increased risk from stroke” (Davis et al., 2012a). Based on a study in a large population of preeclamptic pregnancies, Kajantie et al. (2009) reported that the risk for stroke in subjects born from preeclamptic pregnancies was twice that of controls born from normotensive pregnancies. Other studies have described an increased risk for pulmonary hypertension (Jayet et al., 2010), metabolic and endocrine disease (Wu et al., 2009, 2011), depression (Tuovinen et al., 2010), cerebral palsy (Szymonowicz and Yu, 1987), poor cognitive outcome (Cheng et al., 2004), or intellectual disabilities (Griffith et al., 2011) in children born of preeclamptic pregnancies compared to non-exposed children.

These clinical and epidemiological observations are supported by a recent review of animal models of preeclampsia (Davis et al., 2012b), including those generated by systemic hypoxia, by mechanical reduction of maternal uterine artery blood flow, in genetically modified animals lacking endothelial nitric oxide synthase (eNOS), or by overexpression of sFlt-1 by infection with adenovirus carrying this protein. Notwithstanding differences in design and outcome of these models, the conclusion was that “animal studies support the potential relevance of these insults to programming of offspring blood pressure.”

Although fetal programming by preeclampsia is suggested by human and animal studies, it is not easy to determine whether preeclampsia per se leads to high cardiovascular risk in the offspring or whether related factors, such as intrauterine growth restriction or preterm delivery, contribute. To allay these concerns, most studies have excluded individuals exposed to preterm delivery or intrauterine growth restriction associated with preeclampsia. Interestingly, the risks for hypertension, impaired neurological function, and stroke in offspring from preeclamptic pregnancies remain significant (Kajantie et al., 2009; Tuovinen et al., 2010, 2012). Moreover, a study performed in brothers exposed, or not, to preeclampsia suggested that impaired vessel function was associated with preeclampsia per se rather than genetic predisposition (Jayet et al., 2010). It is plausible, then, that exposure to preeclampsia in utero can predispose to adverse outcomes later in life.

Understanding the underlying mechanisms might suggest interventions to prevent the occurrence of future chronic disease in offspring exposed to preeclampsia. Several groups (Jayet et al., 2010; Lazdam et al., 2010; Kvehaugen et al., 2011; Davis et al., 2012a,b; Lawlor et al., 2012), including ours (Escudero and Sobrevia, 2008; Escudero et al., 2012), have presented evidence of endothelial dysfunction in the feto-placental circulation in preeclampsia, which may be a precursor to the long-term complications observed in offspring born of preeclamptic pregnancies.

Adenosine is a purinergic nucleoside which controls several physiological processes, including angiogenesis and vasculogenesis. Adenosine activates a family of G-coupled adenosine receptors, A₁AR, A_2AAR, A_2BAR, and A₃AR (Olah and Stiles, 2000; Jacobson and Gao, 2006). All of the adenosine receptors have been implicated in the modulation of angiogenesis (see Table 1). Briefly, stimulation of A₁AR on embryonic EPCs promotes their adherence to the vascular endothelium, suggesting an important role for this receptor subtype in vasculogenesis (Ryzhov et al., 2008). A₁AR have also been reported to upregulate VEGF production from monocytes, thus promoting angiogenesis (Clark et al., 2007).

Depending on the tissue or cell studied, A_2AAR and A_2BAR can play a dominant role in the regulation of angiogenic factors. For example, A_2BAR upregulates the pro-angiogenic factors VEGF, basic fibroblast growth factor (bFGF), insulin-like factor-1, and interleukin 8 (IL-8) in human microvascular endothelial cells (Grant et al., 1999; Feoktistov et al., 2002). Conversely, A_2AAR is reported to upregulate VEGF in macrophages (Leibovich et al., 2002; Pinhal-Enfield et al., 2003). Stimulation of A₃AR in mast cells and some tumors can result in upregulation of pro-angiogenic factors, complementing the actions of adenosine mediated via A_2BAR (Feoktistov et al., 2003). Of interest, stimulation of A_2AAR in HMEC-1 inhibits the release of the anti-angiogenic factor thrombospondin 1, providing yet another means by which adenosine may regulate angiogenesis (Desai et al., 2005; see Table 1).

While A_2AAR and A_2BAR have been shown to mediate the proliferative actions of adenosine in human retinal microvascular endothelial cells (Grant et al., 1999, 2001), HUVEC (Feoktistov et al., 2004; Escudero et al., 2012), or porcine coronary artery and rat aortic endothelial cells (Dubey et al., 2002), it remains unclear whether A₁AR and A₃AR are functionally expressed and what role, if any, they play in endothelial cells (Wyatt et al., 2002; Schaddelee et al., 2003).

Although some data suggest that cAMP may play a role in the pro-angiogenic effects of adenosine in certain cells (Takagi et al., 1996), other studies show that upregulation of angiogenic factors is mediated via coupling to Gq, possibly involving mitogen-activated protein kinase (MAPK) pathways (Grant et al., 1999, 2001; Feoktistov et al., 2002; Ryzhov et al., 2014). Further studies, using HMEC-1 demonstrated that adenosine receptor-dependent upregulation of VEGF production was associated with an increase in VEGF transcription, activator protein 1 (AP-1) activity and transcription factor JunB (JunB) accumulation (Ryzhov et al., 2014).

Mechanistically, A_2BAR which are coupled to both Gs and Gq proteins (Feoktistov and Biaggioni, 1995) increase JunB protein levels and VEGF production via stimulation of protein lipase C and extracellular signal-regulated kinase (ERK), which are possibly linked by the calcium diacylglycerol guanine nucleotide exchange factor (CalDAG-GEF)–Ras-proximate-1 (Rap1) pathway (Ryzhov et al., 2014). These effects were protein kinase A (PKA)-independent because the PKA inhibitors had no effect on the A_2BAR-dependent increase in JunB protein levels and VEGF production. Because VEGF secretion and reporter promoter activity induced by the adenosine analog 5′-N-ethylcarboxamido-adenosine (NECA) were inhibited by the expression of a dominant, negative JunB or by JunB knockdown, these data suggest an important role for the A_2B receptor-dependent upregulation of JunB in VEGF production in various cell types, including endothelial cells (Ryzhov et al., 2014).

Another study, in HUVEC, reported that adenosine-mediated activation of ERK may involve an exchange protein activated by cAMP (Epac), a component of a family of cAMP-activated guanine nucleotide exchange factors for Rap GTPases (Fang and Olah, 2007). Thus, A_2BAR, coupled to Gα_s promotes activation of adenylyl cyclase and an increase in intracellular cAMP. In turn, cAMP activates Epac 1, which may then activate a cascade of RapGTPase, B-Raf, and finally, ERK (Fang and Olah, 2007), demonstrating an alternative pathway for ERK activation involved in upregulation of pro-angiogenic proteins.

In addition to JunB, the mediators downstream of ERK and p38 MAPK activation may include molecules such as hypoxia inducible factor type 1 α (HIF-1α) and/or nitric oxide (NO). Using foam cells generated in vitro, Gessi et al. (2010) found that activation of A₃AR, A_2BAR, and to a lesser extent, the A_2A subtypes were associated with the production of VEGF induced by adenosine and hypoxia. This last effect was dependent on activation of ERK, p38 MAPK, Akt, and HIF-1α. Furthermore, adenosine has been reported to increase the synthesis of the angiogenic modulator NO in some, but not all, cultured endothelial cells (Sobrevia et al., 1996; Li et al., 1998; Wyatt et al., 2002). Whether, adenosine receptor-mediated activation of ERK-MAPK-HIF-1α increases NO is unknown. But, it has been reported that NO promotes a regulatory loop with ERK activation/deactivation (Schieke et al., 1999) and stabilization of HIF-1α and promotes HIF-1α binding to DNA (Kimura et al., 2000).

The expression of adenosine receptor subtypes and their function are subject to dynamic regulation by hypoxia (Bshesh et al., 2002; Eltzschig et al., 2003; Feoktistov et al., 2004). Because the A_2BAR promoter contains a functional binding site for HIF-1α (Kong et al., 2006), the onset of hypoxia strongly induces A_2BAR expression in different cell types including human dermal microvascular endothelial cells (Eltzschig et al., 2003), and HUVEC (Feoktistov et al., 2004). In addition, elevated expression of A_2AAR has also been reported after exposure to hypoxia in human placental homogenate (Von Versen-Hoynck et al., 2009), fetal chromaffin-derived cell line (Brown et al., 2011), and human lung endothelial cells, while no evidence of A_2AAR upregulation was seen in mouse endothelial cells (Ahmad et al., 2009).

Interestingly, despite the fact that all adenosine receptors contain a hypoxia response element in their promoters (St. Hilaire et al., 2009), regulation via HIF is differentially modulated. Whereas A_2BAR is regulated by HIF-1α, A_2AAR is regulated by HIF-2α, suggesting that transcriptional regulation might be part of the switch of A_2AAR toward A_2BAR expression observed in HUVEC exposed to hypoxia (Feoktistov et al., 2004). This switch may have important functional implications for regulation of angiogenesis. For example, in HUVEC, adenosine does not stimulate VEGF secretion under normoxic conditions, but hypoxia increases the expression of A_2BAR, which are then able to stimulate VEGF release (Feoktistov et al., 2004). Therefore, we could speculate that switching the expression of adenosine receptor toward A_2BAR rather than A_2AAR during hypoxia in the endothelium may offer some advantages in the angiogenic process, since high levels of adenosine may downregulate activation and/or expression of A_2AAR as mechanisms of desensitization. But at the same time, upregulation of A_2BAR will enhance or maintain the pro-angiogenic capacity of adenosine in conditions were high levels of this autocoid are expected. It is likely that this phenomenon may occur in preeclampsia.

The plasma level of adenosine is finely regulated by a series of enzymes responsible for synthesis and catabolism (see details in Escudero and Sobrevia, 2009). Compared with non-pregnant women, normal, pregnant women exhibit increased synthesis, but reduced catabolism, of adenosine (Yoneyama et al., 2000; Lee et al., 2007). Several studies have described high adenosine levels in both maternal (Yoneyama et al., 2001, 2002a,b,c) and fetal blood (Yoneyama et al., 1996; Espinoza et al., 2011; Escudero and Sobrevia, 2012) during preeclampsia, particularly in severe preeclampsia, compared with normal pregnancies (Yoneyama et al., 1996; Espinoza et al., 2011; Escudero and Sobrevia, 2012). Unexpectedly, these high levels are associated with high adenosine catabolism via adenosine deaminase 2 (ADA2; Yoneyama et al., 2002a; Kafkasli et al., 2006) as well as elevated adenosine uptake (Escudero et al., 2008). The causes and consequences of a high extracellular adenosine level in both maternal and fetal circulation during preeclampsia are unclear; however, it may be explained by an adaptive mechanism (Casanello et al., 2007; Escudero and Sobrevia, 2008, 2009, 2012; Escudero et al., 2009) associated with vasodilation or angiogenesis in preeclamptic placenta as occurs in other tissues, such as heart, muscle, or brain, in unfavorable conditions such as hypoxia (Eckle et al., 2007; Loffler et al., 2007).

The levels of adenosine in umbilical vein blood in preeclampsia (1.7 vs. 0.5 μmol/L, preeclampsia vs. normal pregnancy; Yoneyama et al., 1996; Espinoza et al., 2011) and in the culture medium of human placental microvascular endothelial cells (hPMEC) from preeclamptic pregnancies (2.7 vs. 0.6 μmol/L) are at least three times higher than in normal pregnancy (Escudero et al., 2008), making it likely that, in preeclampsia, all adenosine receptors are likely to be stimulated (Jacobson and Gao, 2006). However, only a few reports have described the effect of preeclampsia on the expression and function of adenosine receptors (Escudero et al., 2008, 2012; Kim et al., 2008; Von Versen-Hoynck et al., 2009). Thus, reduced expression of A_2AAR, without changes in A_2BAR (Escudero et al., 2008), was found in hPMEC isolated from preeclamptic placentas, whereas reduced A_2AAR (Escudero et al., 2012) but higher A_2BAR (Acurio et al., 2014) expression levels were found in HUVEC from preeclampsia. Yet, increased levels of all adenosine receptors have been reported in placental homogenate from preeclamptic placentas compared with normal pregnancy (Von Versen-Hoynck et al., 2009).

It has been shown that activation of A_2AAR leads to reduction in adenosine uptake by the equilibrative nucleotide transporter type 1 (hENT1) and hENT2, whereas A_2BAR activation increases hENT2-mediated adenosine transport in cells from preeclamptic placentas (Escudero et al., 2008). Therefore, during preeclampsia, activation of adenosine receptors may control adenosine transport and, hence, extracellular adenosine levels. However, because adenosine levels are increased despite the elevation of total adenosine uptake, it is expected that the production of adenosine from sources such ATP or cell debris is higher in preeclampsia than in normal pregnancy (Spaans et al., 2014).

Recently, we observed reduced protein abundance of A_2AAR in HUVEC derived from EOPE, but non-significant changes in LOPE, compared with cells from normal pregnancy. These findings were associated with a basal (i.e., without any treatment) reduction in cell migration/proliferation of HUVEC in EOPE compared with normal pregnancy or LOPE (Escudero et al., 2012). In addition, CGS-21680 (an A_2AAR agonist) and NECA significantly increased HUVEC migration/proliferation in normal pregnancy, LOPE, and EOPE. However, considering that cells from EOPE exhibited the lowest migration/proliferation in the basal conditions, the magnitude of response to both adenosine receptor agonists in migration and proliferation tends to be higher in cells from EOPE than those from others groups. In agreement with these results, VEGF expression was significantly lower in HUVEC from EOPE, but higher in LOPE, compared to normal pregnancy. Also, CGS-21680 increases the protein abundance of VEGF in normal and EOPE-derived cells, an effect blocked by the A_2AAR antagonist ZM-241385. Nevertheless, CGS-21680 did not affect VEGF expression in HUVEC from LOPE, but ZM-241385 led to a reduction (41 ± 6%, p < 0.05) in the level of this protein compared to corresponding levels at basal condition, suggesting that A_2AAR is activated at basal condition in LOPE. Thus, A_2AAR-mediated HUVEC proliferation and migration was associated with VEGF synthesis in normal pregnancy, LOPE, and EOPE.

To elucidate potential intracellular pathways related to A_2AAR activation, we determined that CGS-21680 increased the synthesis of NO as evidenced by activation of eNOS (i.e., the p-eNOS/eNOS ratio) and nitrite and nitrotyrosine levels in HUVEC from normal pregnancies and EOPE, but not in LOPE. The stimulatory effect observed in normal and EOPE cells was blocked by ZM-241385 co-incubation. In contrast, ZM-241385 reduced NO synthesis in LOPE cells compared to non-treated controls. Furthermore, using the non-selective nitric oxide synthase inhibitor, L-NAME, we found a significant reduction in the HUVEC migration/proliferation responses and VEGF protein levels in cells from normal pregnancies and LOPE, but not in EOPE cells stimulated with CGS-21680.

Thus, our study demonstrated that activation of A_2AAR is associated with the following cascade: eNOS activation (i.e., ser¹¹⁷⁷ phosphorylation), NO synthesis, nitrotyrosine formation, VEGF expression, and cell proliferation/migration in normal pregnancy. However, cells derived from EOPE and LOPE were different in several aspects. Whereas EOPE cells exhibited low A_2AAR expression and reduction of NO/VEGF synthesis and cell proliferation/migration; LOPE cells demonstrated increased cell proliferation/migration, mediated in part through the same pathway (see Figure 2). Existence of this pathway was recently confirmed using selective shRNA for A_2AAR in HUVEC. Knockdown of A_2AAR was associated with reduced formation of intracellular cAMP, NO metabolites, VEGF protein level, and the capacity for tube formation compared with controls (unpublished results).

As stated before, adenosine-dependent angiogenesis can be regulated by all four adenosine receptors. There is little information on the role of A_2BAR in the angiogenic process during preeclampsia, whereas the participation of A₁AR and A₃AR in this process is unknown. In primary cultures of hPMEC, a cell type with high pro-angiogenic capacity compared to HUVEC (Dye et al., 2004), we found that A_2BAR may be constitutively activated in cells from preeclamptic placentas, since the use of A_2A/A_2BAR inhibitors in non-stimulated cells decreases adenosine uptake (Escudero et al., 2008). More recently, we found that activation of A_2BAR in HUVEC accounts for at least 30% of the pro-proliferative response mediated by adenosine or NECA (Acurio et al., 2014). These data agree with prior reports (Feoktistov et al., 2002, 2004) that exposure of HUVEC to hypoxia increases the expression of A_2BAR, which is then able to stimulate VEGF release.

In view of available information, we can speculate that during preeclampsia, a condition associated with reduction in the expression and activity of A_2AAR, a compensatory increase in the expression and/or activity of A_2BAR occurs that tends to compensate the impaired adenosine-mediated pro-angiogenic process. Moreover, since adenosine is pro-angiogenic, the reduction in A_2AAR expression and down activation of A_2AAR-dependent intracellular pathway might be part of the apparent “adenosine paradox,” in which increased adenosine levels do not stimulate angiogenesis in preeclampsia. The mechanism underlying this phenomenon is unclear, but may be associated with the capacity of adenosine to regulate the expression of its receptors, as exhibited in cells such as cardiomyocytes (Headrick et al., 2013) or PC12 cells (Saitoh et al., 1994). In particular, PC12 exposure to A_2AAR agonists reduces ADORA2 gene expression (Saitoh et al., 1994), suggesting a transcriptional regulation of A_2AAR by adenosine. Whether similar regulation is present in endothelium is unknown, but this is a possible mechanism for the pathophysiological deregulation observed in preeclampsia.

Although the intracellular signaling pathway related to adenosine receptor activation is an area of active research, only our recent study suggested a potential adenosine receptor-dependent mechanism in preeclampsia. On the basis of this study, we propose a model (Figure 2), in which low expression of A_2AAR in EOPE leads to reduction in NO and VEGF expression (Escudero et al., 2012). The implication of these alterations for feto-placental angiogenesis is poorly understood, but might involve changes in the activation of HIF (Kimura et al., 2002) and changes in the promoter activity of several proteins, including anti-angiogenic factors such as thrombospondin 2 (MacLauchlan et al., 2011) or pro-angiogenic factors like VEGF (Kimura et al., 2002; Feoktistov et al., 2004). Considering that NO can cause nitration of tyrosine residues on HIF-1α (Riano et al., 2011), and may contribute to stabilization, we propose that the reduced adenosine-mediated NO synthesis observed in EOPE might be associated with impaired HIF-dependent VEGF expression (see Figure 3). Clearly, more studies are necessary to understand all the processes involved in these alterations.

Another question that needs to be answered is whether impaired adenosine-mediated angiogenesis in the feto-placental circulation of preeclamptic pregnancies persists after birth. In this context, Coney and Marshall (2010) have demonstrated that prenatal hypoxia has long-lasting effects on vascular function in the skeletal muscle of adult male rats. In particular, in a group of adult males, they investigated how chronic systemic hypoxia in utero (CHU) affects the cardiovascular response evoked by acute, systemic hypoxia. One of the most intriguing results was the fact that the overall magnitude of vasodilator response evoked in muscle by acute systemic hypoxia is similar in CHU and normoxic rats, but the mechanisms underlying the response appear to be different. Thus, they conclude that, whereas in normoxia, vasodilatory response is associated with the activation of endothelial A₁AR and NO-dependent effects, in CHU, participation of A₁AR is limited, and vasodilatory response in the muscle is replaced by factors other than adenosine. Moreover, it has been reported that mice deficient in A_2AAR (KO-A_2AAR) exhibit no significant difference in systemic blood pressure compared to wild-type animals, but they do develop pulmonary artery hypertension and pulmonary vascular remodeling (Xu et al., 2011). Thus, these studies demonstrate how the adenosine-impaired angiogenesis and vascular remodeling observed in pathological pregnancies such as preeclampsia may be related to future cardiovascular risk.